This is a very "raw" package, so it probably will not install as smoothly as one might hope. This is partially a complexity with Octave and g++. It is also because I don't know enough about Octave conventions and styles. (I would be grateful to any Octave users who would like to suggest changes to make this look more like an Octave package.)
Build R as a shared library
On Unix, this is done by configuring and building R as follows
 ./configure --enable-R-shlib
 make
This creates $R_HOME/bin/libR.so which we will link against.
Compile the Octave code in this package.
If you get errors in this step, read the FAQ.html file. If there is no answer there, let me know.

Running Octave and R

Currently, to access the R facilities from within Octave, you will need to specify some environment variables before running Octave.
Add the directory containing the R shared library (libR.so) to the LD_LIBRARY_PATH. This is done with the command
      setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:`R RHOME`/bin
if you are using the Csh ro Tcsh shell, or as
      LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:`R RHOME`/bin
      export LD_LIBRARY_PATH
if you are using some form of the Bourne shell (e.g. sh, bash, etc.)
You will also need to add the directory in which you built the ROctave code to the LD_LIBRARY_PATH.
This is needed to locate the libConverters.so file.
Add this same directory (containing the .oct files) to the environment variable OCTAVE_PATH
At this point you should be able to invoke octave and call an R function.
% octave
octave:1> ROctave_callR("rnorm", 10)
ans =

   0.527109
   1.408802
  -1.789294
   1.378379
   1.066724
   0.083658
   0.712621
   0.078647
   0.842873
  -2.706941

Duncan Temple Lang <duncan@research.bell-labs.com>
Last modified: Thu Oct 10 14:49:55 EDT 2002